pyannote.audio segments audio by speaker, answering "who spoke when" — the missing piece that turns a transcript into a usable meeting record.
| Category | Speech (STT / TTS) |
| Type | Speaker diarization |
| License | MIT |
| Runs locally | Yes |
| Built with | Python |
| Skill level | Intermediate |
| Best for | meeting transcripts with several speakers |
Other open-source speech (stt / tts) tools worth comparing:
WhisperOpenAI's open speech-to-text baseline
faster-whisperWhisper, much faster and lighter
WhisperXWhisper plus word timestamps and diarization
PiperFast, local neural text-to-speech
Coqui TTSDeep-learning TTS with voice cloning
BarkText-to-audio with voices and effects
F5-TTSZero-shot voice cloning that actually convinces
KokoroTiny 82M TTS with astonishing quality
whisper.cppWhisper in pure C/C++, runs anywhere
OpenVoiceClone a voice and control its emotion
StyleTTS 2Human-level speech synthesis
Silero VADDetect speech in audio, instantlypyannote.audio is free and open-source (MIT license), so you can use, self-host and modify it at no cost.
Yes. pyannote.audio is designed to run on your own machine or server, keeping your data private.
Popular open-source alternatives include Whisper, faster-whisper, WhisperX. See the comparisons above to choose.
Browse the full directory of open-source AI tools, models and projects — updated daily.
Browse all tools →